Adding Name and ISeekable improvements (#177)

docs/parsers.md

@@ -553,7 +553,7 @@ Result:
 
 ### Deferred
 
-Creates a parser that can be references before it is actually defined. This is used when there is a cyclic dependency between parsers.
+Creates a parser that can be referenced before it is actually defined. This is used when there is a cyclic dependency between parsers.
 
 ```c#
 Deferred<T> Deferred<T>()
@@ -777,6 +777,7 @@ Result:
 ```
 failure: "Unexpected char c"
 ```
+
 ### When
 
 Adds some additional logic for a parser to succeed.
@@ -823,6 +824,15 @@ Parser<U> Discard<U>()
 Parser<U> Discard<U>(U value)
 ```
 
+### Lookup
+
+Builds a parser that lists all possible matches to improve performance. Most parsers implement `ISeekable` parsers in order to provide `OneOf` a way to build a lookup table and identify the potential next parsers in the chain. Some parsers don't implement `ISeekable` because they are built too late, like `Deferred`. The `Lookup` parser circumvents that lack.
+
+```c#
+Parser<T> Lookup<U>(params ReadOnlySpan<char> expectedChars)
+Parser<T> Lookup(params ISeekable[] parsers)
+```
+
 ## Other parsers
 
 ### AnyCharBefore

docs/writing.md

@@ -25,12 +25,15 @@ public override bool Parse(ParseContext context, ref ParseResult<ValueTuple<T1,
         if (_parser2.Parse(context, ref parseResult2))
         {
             result.Set(parseResult1.Start, parseResult2.End, new ValueTuple<T1, T2>(parseResult1.Value, parseResult2.Value));
+
+            context.ExitParser(this);
             return true;
         }
 
         context.Scanner.Cursor.ResetPosition(start);
     }
 
+    context.ExitParser(this);
     return false;
 }
 ```
@@ -51,10 +54,12 @@ public override bool Parse(ParseContext context, ref ParseResult<T> result)
     {
         if (parser.Parse(context, ref result))
         {
+            context.ExitParser(this);
             return true;
         }
     }
 
+    context.ExitParser(this);
     return false;
 }
 ```

src/Parlot/CharMap.cs

@@ -1,12 +1,19 @@
-using System;
+using System;
 using System.Collections.Generic;
 using System.Linq;
+using System.Reflection;
 using System.Runtime.CompilerServices;
 
 namespace Parlot;
 
+/// <summary>
+/// Combines maps of ascii and non-ascii characters.
+/// If all characters are ascii, the non-ascii dictionary is not used.
+/// </summary>
 internal sealed class CharMap<T> where T : class
 {
+    public static MethodInfo IndexerMethodInfo = typeof(CharMap<T>).GetMethod("get_Item", BindingFlags.Public | BindingFlags.Instance)!;
+
     private readonly T[] _asciiMap = new T[128];
     private Dictionary<uint, T>? _nonAsciiMap;
 

src/Parlot/Compilation/CompilationContext.cs

@@ -1,4 +1,4 @@
-using Parlot.Fluent;
+using Parlot.Fluent;
 using System;
 using System.Collections.Generic;
 using System.Linq.Expressions;
@@ -48,7 +48,7 @@ public CompilationContext()
     /// Gets or sets whether the current compilation phase should ignore the results of the parsers.
     /// </summary>
     /// <remarks>
-    /// When set to false, the compiled statements don't need to record and define the <see cref="CompilationResult.Value"/> property.
+    /// When set to true, the compiled statements don't need to record and define the <see cref="CompilationResult.Value"/> property.
     /// This is done to optimize compiled parser that are usually used for pattern matching only.
     /// </remarks>
     public bool DiscardResult { get; set; }
@@ -85,7 +85,12 @@ public CompilationResult CreateCompilationResult(Type valueType, bool defaultSuc
         result.Variables.Add(valueVariable);
 
         result.Body.Add(Expression.Assign(successVariable, Expression.Constant(defaultSuccess, typeof(bool))));
-        result.Body.Add(Expression.Assign(valueVariable, defaultValue ?? Expression.Default(valueType)));
+
+        // Don't need to assign a type's default value
+        if (defaultValue != null)
+        {
+            result.Body.Add(Expression.Assign(valueVariable, defaultValue ?? Expression.Default(valueType)));
+        }
 
         return result;
     }

src/Parlot/Compilation/CompiledParser.cs

@@ -1,4 +1,4 @@
-using Parlot.Fluent;
+using Parlot.Fluent;
 using System;
 
 namespace Parlot.Compilation;
@@ -26,22 +26,28 @@ public class CompiledParser<T> : Parser<T>, ICompiledParser
 
     public CompiledParser(Func<ParseContext, ValueTuple<bool, T>> parse, Parser<T> source)
     {
+        Name = "Compiled";
         _parse = parse ?? throw new ArgumentNullException(nameof(parse));
         Source = source;
     }
 
     public override bool Parse(ParseContext context, ref ParseResult<T> result)
     {
+        context.EnterParser(this);
+
         var cursor = context.Scanner.Cursor;
         var start = cursor.Offset;
         var parsed = _parse(context);
 
         if (parsed.Item1)
         {
             result.Set(start, cursor.Offset, parsed.Item2);
+
+            context.ExitParser(this);
             return true;
         }
 
+        context.ExitParser(this);
         return false;
     }
 }

src/Parlot/Compilation/ExpressionHelper.cs

@@ -1,4 +1,4 @@
-using Parlot.Fluent;
+using Parlot.Fluent;
 using System;
 using System.Linq.Expressions;
 using System.Reflection;
@@ -8,7 +8,7 @@ namespace Parlot.Compilation;
 
 public static class ExpressionHelper
 {
-    internal static readonly MethodInfo ParserContext_SkipWhiteSpaceMethod = typeof(ParseContext).GetMethod(nameof(ParseContext.SkipWhiteSpace), Array.Empty<Type>())!;
+    internal static readonly MethodInfo ParserContext_SkipWhiteSpaceMethod = typeof(ParseContext).GetMethod(nameof(ParseContext.SkipWhiteSpace), [])!;
     internal static readonly MethodInfo ParserContext_WhiteSpaceParser = typeof(ParseContext).GetProperty(nameof(ParseContext.WhiteSpaceParser))?.GetGetMethod()!;
     internal static readonly MethodInfo Scanner_ReadText_NoResult = typeof(Scanner).GetMethod(nameof(Parlot.Scanner.ReadText), [typeof(ReadOnlySpan<char>), typeof(StringComparison)])!;
     internal static readonly MethodInfo Scanner_ReadChar = typeof(Scanner).GetMethod(nameof(Parlot.Scanner.ReadChar), [typeof(char)])!;
@@ -23,23 +23,21 @@ public static class ExpressionHelper
     internal static readonly MethodInfo Scanner_ReadDoubleQuotedString = typeof(Scanner).GetMethod(nameof(Parlot.Scanner.ReadDoubleQuotedString), [])!;
     internal static readonly MethodInfo Scanner_ReadQuotedString = typeof(Scanner).GetMethod(nameof(Parlot.Scanner.ReadQuotedString), [])!;
 
-    internal static readonly MethodInfo Cursor_Advance = typeof(Cursor).GetMethod(nameof(Parlot.Cursor.Advance), Array.Empty<Type>())!;
+    internal static readonly MethodInfo Cursor_Advance = typeof(Cursor).GetMethod(nameof(Parlot.Cursor.Advance), [])!;
     internal static readonly MethodInfo Cursor_AdvanceNoNewLines = typeof(Cursor).GetMethod(nameof(Parlot.Cursor.AdvanceNoNewLines), [typeof(int)])!;
     internal static readonly MethodInfo Cursor_ResetPosition = typeof(Cursor).GetMethod("ResetPosition")!;
+    internal static readonly MethodInfo Character_IsWhiteSpace = typeof(Character).GetMethod(nameof(Character.IsWhiteSpace))!;
 
     internal static readonly ConstructorInfo Exception_ToString = typeof(Exception).GetConstructor([typeof(string)])!;
 
     internal static readonly ConstructorInfo TextSpan_Constructor = typeof(TextSpan).GetConstructor([typeof(string), typeof(int), typeof(int)])!;
 
     internal static readonly MethodInfo ReadOnlySpan_ToString = typeof(ReadOnlySpan<char>).GetMethod(nameof(ToString), [])!;
-    
+
     internal static readonly MethodInfo MemoryExtensions_AsSpan = typeof(MemoryExtensions).GetMethod(nameof(MemoryExtensions.AsSpan), [typeof(string)])!;
-
     public static Expression ArrayEmpty<T>() => ((Expression<Func<object>>)(() => Array.Empty<T>())).Body;
     public static Expression New<T>() where T : new() => ((Expression<Func<T>>)(() => new T())).Body;
 
-    public static readonly Expression<Func<Cursor, char, char, bool>> CharacterIsInRange = (cursor, b, c) => Character.IsInRange(cursor.Current, b, c);
-
     //public static Expression NewOptionalResult<T>(this CompilationContext _, Expression hasValue, Expression value) => Expression.New(GetOptionalResult_Constructor<T>(), [hasValue, value]);
     public static Expression NewTextSpan(this CompilationContext _, Expression buffer, Expression offset, Expression count) => Expression.New(TextSpan_Constructor, [buffer, offset, count]);
     public static MemberExpression Scanner(this CompilationContext context) => Expression.Field(context.ParseContext, "Scanner");
@@ -53,11 +51,28 @@ public static class ExpressionHelper
     public static MemberExpression Buffer(this CompilationContext context) => Expression.Field(context.Scanner(), "Buffer");
     public static Expression ThrowObject(this CompilationContext _, Expression o) => Expression.Throw(Expression.New(Exception_ToString, Expression.Call(o, o.Type.GetMethod("ToString", [])!)));
     public static Expression ThrowParseException(this CompilationContext context, Expression message) => Expression.Throw(Expression.New(typeof(ParseException).GetConstructors().First(), [message, context.Position()]));
+    public static Expression BreakPoint(this CompilationContext _, Expression state, Action<object> action) => Expression.Invoke(Expression.Constant(action, typeof(Action<object>)), Expression.Convert(state, typeof(object)));
 
     public static MethodCallExpression ReadSingleQuotedString(this CompilationContext context) => Expression.Call(context.Scanner(), Scanner_ReadSingleQuotedString);
     public static MethodCallExpression ReadDoubleQuotedString(this CompilationContext context) => Expression.Call(context.Scanner(), Scanner_ReadDoubleQuotedString);
     public static MethodCallExpression ReadQuotedString(this CompilationContext context) => Expression.Call(context.Scanner(), Scanner_ReadQuotedString);
     public static MethodCallExpression ReadChar(this CompilationContext context, char c) => Expression.Call(context.Scanner(), Scanner_ReadChar, Expression.Constant(c));
+
+    // Surprisingly whiting the same direct code with this helper is slower that calling scanner.ReadChar()
+    public static Expression ReadCharInlined(this CompilationContext context, char c, CompilationResult result)
+    {
+        var constant = Expression.Constant(c);
+        return Expression.IfThen(
+            Expression.Equal(context.Current(), constant), // if (cursor.Current == 'c')
+            Expression.Block(
+                Expression.Assign(result.Success, TrueExpression),
+                context.Advance(),
+                context.DiscardResult
+                    ? Expression.Empty()
+                    : Expression.Assign(result.Value, constant)
+                )
+            );
+    }
     public static MethodCallExpression ReadDecimal(this CompilationContext context) => Expression.Call(context.Scanner(), Scanner_ReadDecimal);
     public static MethodCallExpression ReadDecimal(this CompilationContext context, Expression allowLeadingSign, Expression allowDecimalSeparator, Expression allowGroupSeparator, Expression allowExponent, Expression number, Expression decimalSeparator, Expression groupSeparator) => Expression.Call(context.Scanner(), Scanner_ReadDecimalAllArguments, allowLeadingSign, allowDecimalSeparator, allowGroupSeparator, allowExponent, number, decimalSeparator, groupSeparator);
     public static MethodCallExpression ReadInteger(this CompilationContext context) => Expression.Call(context.Scanner(), Scanner_ReadInteger);
@@ -88,4 +103,6 @@ public static MethodCallExpression ParserSkipWhiteSpace(this CompilationContext
     {
         return Expression.Call(context.ParseContext, ParserContext_SkipWhiteSpaceMethod);
     }
+
+    public static ConstantExpression TrueExpression { get; } = Expression.Constant(true, typeof(bool));
 }

src/Parlot/Fluent/Always.cs

@@ -1,4 +1,4 @@
-using Parlot.Compilation;
+using Parlot.Compilation;
 using System.Linq.Expressions;
 
 namespace Parlot.Fluent;
@@ -12,6 +12,7 @@ public sealed class Always<T> : Parser<T>, ICompilable
 
     public Always(T value)
     {
+        Name = "Always";
         _value = value;
     }
 
@@ -21,6 +22,7 @@ public override bool Parse(ParseContext context, ref ParseResult<T> result)
 
         result.Set(context.Scanner.Cursor.Offset, context.Scanner.Cursor.Offset, _value);
 
+        context.ExitParser(this);
         return true;
     }
 

src/Parlot/Fluent/Between.cs

@@ -1,4 +1,4 @@
-using Parlot.Compilation;
+using Parlot.Compilation;
 using Parlot.Rewriting;
 using System;
 using System.Linq.Expressions;
@@ -23,6 +23,8 @@ public Between(Parser<A> before, Parser<T> parser, Parser<B> after)
             ExpectedChars = seekable.ExpectedChars;
             SkipWhitespace = seekable.SkipWhitespace;
         }
+
+        Name = $"Between({before.Name},{parser.Name},{after.Name})";
     }
 
     public bool CanSeek { get; }
@@ -43,13 +45,17 @@ public override bool Parse(ParseContext context, ref ParseResult<T> result)
 
         if (!_before.Parse(context, ref parsedA))
         {
+            context.ExitParser(this);
+
             // Don't reset position since _before should do it
             return false;
         }
 
         if (!_parser.Parse(context, ref result))
         {
             cursor.ResetPosition(start);
+
+            context.ExitParser(this);
             return false;
         }
 
@@ -58,9 +64,12 @@ public override bool Parse(ParseContext context, ref ParseResult<T> result)
         if (!_after.Parse(context, ref parsedB))
         {
             cursor.ResetPosition(start);
+
+            context.ExitParser(this);
             return false;
         }
 
+        context.ExitParser(this);
         return true;
     }
 

src/Parlot/Fluent/Capture.cs

@@ -1,4 +1,4 @@
-using Parlot.Compilation;
+using Parlot.Compilation;
 using System.Linq.Expressions;
 
 namespace Parlot.Fluent;
@@ -10,6 +10,7 @@ public sealed class Capture<T> : Parser<TextSpan>, ICompilable
     public Capture(Parser<T> parser)
     {
         _parser = parser;
+        Name = $"{parser.Name} (Capture)";
     }
 
     public override bool Parse(ParseContext context, ref ParseResult<TextSpan> result)
@@ -28,11 +29,11 @@ public override bool Parse(ParseContext context, ref ParseResult<TextSpan> resul
 
             result.Set(start.Offset, end, new TextSpan(context.Scanner.Buffer, start.Offset, length));
 
+            context.ExitParser(this);
             return true;
         }
 
-        context.Scanner.Cursor.ResetPosition(start);
-
+        context.ExitParser(this);
         return false;
     }
 
@@ -61,31 +62,25 @@ public CompilationResult Compile(CompilationContext context)
         //   
         //     success = true;
         // }
-        // else
-        // {
-        //     context.Scanner.Cursor.ResetPosition(start);
-        // }
 
-        var startOffset = context.Offset(start);
+        var startOffset = result.DeclareVariable<int>($"startOffset{context.NextNumber}", context.Offset(start));
 
         result.Body.Add(
             Expression.Block(
                 parserCompileResult.Variables,
                 Expression.Block(parserCompileResult.Body),
-                Expression.IfThenElse(
-                    parserCompileResult.Success,
-                    Expression.Block(
-                        context.DiscardResult
-                        ? Expression.Empty()
-                        : Expression.Assign(result.Value,
+                Expression.IfThen(
+                    test: parserCompileResult.Success,
+                    ifTrue: Expression.Block(
+                        // Never discard result here, that would nullify this parser
+                        Expression.Assign(result.Value,
                             context.NewTextSpan(
                                 context.Buffer(),
                                 startOffset,
                                 Expression.Subtract(context.Offset(), startOffset)
                                 )),
                         Expression.Assign(result.Success, Expression.Constant(true, typeof(bool)))
-                        ),
-                    context.ResetPosition(start)
+                        )
                 )
             )
         );

src/Parlot/Fluent/CharLiteral.cs

@@ -1,4 +1,4 @@
-using Parlot.Compilation;
+using Parlot.Compilation;
 using Parlot.Rewriting;
 using System.Linq.Expressions;
 
@@ -10,6 +10,7 @@ public CharLiteral(char c)
     {
         Char = c;
         ExpectedChars = [c];
+        Name = $"Char('{c}')";
     }
 
     public char Char { get; }
@@ -31,9 +32,12 @@ public override bool Parse(ParseContext context, ref ParseResult<char> result)
             var start = cursor.Offset;
             cursor.Advance();
             result.Set(start, cursor.Offset, Char);
+
+            context.ExitParser(this);
             return true;
         }
 
+        context.ExitParser(this);
         return false;
     }